e2e_pg_loss.py 9.4 KB
Newer Older
1
# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve.
J
Jethong 已提交
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#    http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

from paddle import nn
import paddle
import numpy as np
import copy

24
from .det_basic_loss import DiceLoss
J
Jethong 已提交
25 26 27


class PGLoss(nn.Layer):
28
    def __init__(self, eps=1e-6, **kwargs):
J
Jethong 已提交
29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80
        super(PGLoss, self).__init__()
        self.dice_loss = DiceLoss(eps=eps)

    def org_tcl_rois(self, batch_size, pos_lists, pos_masks, label_lists):
        """
        """
        pos_lists_, pos_masks_, label_lists_ = [], [], []
        img_bs = batch_size
        tcl_bs = 64
        ngpu = int(batch_size / img_bs)
        img_ids = np.array(pos_lists, dtype=np.int32)[:, 0, 0].copy()
        pos_lists_split, pos_masks_split, label_lists_split = [], [], []
        for i in range(ngpu):
            pos_lists_split.append([])
            pos_masks_split.append([])
            label_lists_split.append([])

        for i in range(img_ids.shape[0]):
            img_id = img_ids[i]
            gpu_id = int(img_id / img_bs)
            img_id = img_id % img_bs
            pos_list = pos_lists[i].copy()
            pos_list[:, 0] = img_id
            pos_lists_split[gpu_id].append(pos_list)
            pos_masks_split[gpu_id].append(pos_masks[i].copy())
            label_lists_split[gpu_id].append(copy.deepcopy(label_lists[i]))
        # repeat or delete
        for i in range(ngpu):
            vp_len = len(pos_lists_split[i])
            if vp_len <= tcl_bs:
                for j in range(0, tcl_bs - vp_len):
                    pos_list = pos_lists_split[i][j].copy()
                    pos_lists_split[i].append(pos_list)
                    pos_mask = pos_masks_split[i][j].copy()
                    pos_masks_split[i].append(pos_mask)
                    label_list = copy.deepcopy(label_lists_split[i][j])
                    label_lists_split[i].append(label_list)
            else:
                for j in range(0, vp_len - tcl_bs):
                    c_len = len(pos_lists_split[i])
                    pop_id = np.random.permutation(c_len)[0]
                    pos_lists_split[i].pop(pop_id)
                    pos_masks_split[i].pop(pop_id)
                    label_lists_split[i].pop(pop_id)
        # merge
        for i in range(ngpu):
            pos_lists_.extend(pos_lists_split[i])
            pos_masks_.extend(pos_masks_split[i])
            label_lists_.extend(label_lists_split[i])
        return pos_lists_, pos_masks_, label_lists_

    def pre_process(self, label_list, pos_list, pos_mask):
81 82 83
        max_len = 30  # the max texts in a single image
        max_str_len = 50  # the max len in a single text
        pad_num = 36  # padding num
J
Jethong 已提交
84
        label_list = label_list.numpy()
85
        batch, _, _, _ = label_list.shape
J
Jethong 已提交
86 87 88 89 90
        pos_list = pos_list.numpy()
        pos_mask = pos_mask.numpy()
        pos_list_t = []
        pos_mask_t = []
        label_list_t = []
91 92
        for i in range(batch):
            for j in range(max_len):
J
Jethong 已提交
93 94 95 96 97
                if pos_mask[i, j].any():
                    pos_list_t.append(pos_list[i][j])
                    pos_mask_t.append(pos_mask[i][j])
                    label_list_t.append(label_list[i][j])
        pos_list, pos_mask, label_list = self.org_tcl_rois(
98
            batch, pos_list_t, pos_mask_t, label_list_t)
J
Jethong 已提交
99 100
        label = []
        tt = [l.tolist() for l in label_list]
101
        for i in range(batch):
J
Jethong 已提交
102
            k = 0
103 104
            for j in range(max_str_len):
                if tt[i][j][0] != pad_num:
J
Jethong 已提交
105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214
                    k += 1
                else:
                    break
            label.append(k)
        label = paddle.to_tensor(label)
        label = paddle.cast(label, dtype='int64')
        pos_list = paddle.to_tensor(pos_list)
        pos_mask = paddle.to_tensor(pos_mask)
        label_list = paddle.squeeze(paddle.to_tensor(label_list), axis=2)
        label_list = paddle.cast(label_list, dtype='int32')
        return pos_list, pos_mask, label_list, label

    def border_loss(self, f_border, l_border, l_score, l_mask):
        l_border_split, l_border_norm = paddle.tensor.split(
            l_border, num_or_sections=[4, 1], axis=1)
        f_border_split = f_border
        b, c, h, w = l_border_norm.shape
        l_border_norm_split = paddle.expand(
            x=l_border_norm, shape=[b, 4 * c, h, w])
        b, c, h, w = l_score.shape
        l_border_score = paddle.expand(x=l_score, shape=[b, 4 * c, h, w])
        b, c, h, w = l_mask.shape
        l_border_mask = paddle.expand(x=l_mask, shape=[b, 4 * c, h, w])
        border_diff = l_border_split - f_border_split
        abs_border_diff = paddle.abs(border_diff)
        border_sign = abs_border_diff < 1.0
        border_sign = paddle.cast(border_sign, dtype='float32')
        border_sign.stop_gradient = True
        border_in_loss = 0.5 * abs_border_diff * abs_border_diff * border_sign + \
                         (abs_border_diff - 0.5) * (1.0 - border_sign)
        border_out_loss = l_border_norm_split * border_in_loss
        border_loss = paddle.sum(border_out_loss * l_border_score * l_border_mask) / \
                      (paddle.sum(l_border_score * l_border_mask) + 1e-5)
        return border_loss

    def direction_loss(self, f_direction, l_direction, l_score, l_mask):
        l_direction_split, l_direction_norm = paddle.tensor.split(
            l_direction, num_or_sections=[2, 1], axis=1)
        f_direction_split = f_direction
        b, c, h, w = l_direction_norm.shape
        l_direction_norm_split = paddle.expand(
            x=l_direction_norm, shape=[b, 2 * c, h, w])
        b, c, h, w = l_score.shape
        l_direction_score = paddle.expand(x=l_score, shape=[b, 2 * c, h, w])
        b, c, h, w = l_mask.shape
        l_direction_mask = paddle.expand(x=l_mask, shape=[b, 2 * c, h, w])
        direction_diff = l_direction_split - f_direction_split
        abs_direction_diff = paddle.abs(direction_diff)
        direction_sign = abs_direction_diff < 1.0
        direction_sign = paddle.cast(direction_sign, dtype='float32')
        direction_sign.stop_gradient = True
        direction_in_loss = 0.5 * abs_direction_diff * abs_direction_diff * direction_sign + \
                            (abs_direction_diff - 0.5) * (1.0 - direction_sign)
        direction_out_loss = l_direction_norm_split * direction_in_loss
        direction_loss = paddle.sum(direction_out_loss * l_direction_score * l_direction_mask) / \
                         (paddle.sum(l_direction_score * l_direction_mask) + 1e-5)
        return direction_loss

    def ctcloss(self, f_char, tcl_pos, tcl_mask, tcl_label, label_t):
        f_char = paddle.transpose(f_char, [0, 2, 3, 1])
        tcl_pos = paddle.reshape(tcl_pos, [-1, 3])
        tcl_pos = paddle.cast(tcl_pos, dtype=int)
        f_tcl_char = paddle.gather_nd(f_char, tcl_pos)
        f_tcl_char = paddle.reshape(f_tcl_char,
                                    [-1, 64, 37])  # len(Lexicon_Table)+1
        f_tcl_char_fg, f_tcl_char_bg = paddle.split(f_tcl_char, [36, 1], axis=2)
        f_tcl_char_bg = f_tcl_char_bg * tcl_mask + (1.0 - tcl_mask) * 20.0
        b, c, l = tcl_mask.shape
        tcl_mask_fg = paddle.expand(x=tcl_mask, shape=[b, c, 36 * l])
        tcl_mask_fg.stop_gradient = True
        f_tcl_char_fg = f_tcl_char_fg * tcl_mask_fg + (1.0 - tcl_mask_fg) * (
            -20.0)
        f_tcl_char_mask = paddle.concat([f_tcl_char_fg, f_tcl_char_bg], axis=2)
        f_tcl_char_ld = paddle.transpose(f_tcl_char_mask, (1, 0, 2))
        N, B, _ = f_tcl_char_ld.shape
        input_lengths = paddle.to_tensor([N] * B, dtype='int64')
        cost = paddle.nn.functional.ctc_loss(
            log_probs=f_tcl_char_ld,
            labels=tcl_label,
            input_lengths=input_lengths,
            label_lengths=label_t,
            blank=36,
            reduction='none')
        cost = cost.mean()
        return cost

    def forward(self, predicts, labels):
        images, tcl_maps, tcl_label_maps, border_maps \
            , direction_maps, training_masks, label_list, pos_list, pos_mask = labels
        # for all the batch_size
        pos_list, pos_mask, label_list, label_t = self.pre_process(
            label_list, pos_list, pos_mask)

        f_score, f_boder, f_direction, f_char = predicts
        score_loss = self.dice_loss(f_score, tcl_maps, training_masks)
        border_loss = self.border_loss(f_boder, border_maps, tcl_maps,
                                       training_masks)
        direction_loss = self.direction_loss(f_direction, direction_maps,
                                             tcl_maps, training_masks)
        ctc_loss = self.ctcloss(f_char, pos_list, pos_mask, label_list, label_t)
        loss_all = score_loss + border_loss + direction_loss + 5 * ctc_loss

        losses = {
            'loss': loss_all,
            "score_loss": score_loss,
            "border_loss": border_loss,
            "direction_loss": direction_loss,
            "ctc_loss": ctc_loss
        }
        return losses